Deriving Differential Equations from the Riccati Equation for Optimal Control

Click For Summary
SUMMARY

The discussion focuses on deriving three separate differential equations from the Riccati equation in the context of a Linear Regulator Problem for Optimal Control, as outlined in Donald Kirk's book "Optimal Control." The user seeks clarification on how a single matrix equation can lead to multiple differential equations. It is established that the symmetric nature of the matrix K leads to four coupled differential equations, of which only three are unique due to symmetry. The user ultimately gains clarity on the relationship between the matrices involved.

PREREQUISITES
  • Understanding of Riccati equations in control theory
  • Familiarity with MATLAB for numerical computations
  • Knowledge of linear algebra, particularly matrix operations
  • Basic concepts of optimal control theory as presented in "Optimal Control" by Donald Kirk
NEXT STEPS
  • Study the derivation of the Riccati equation in optimal control problems
  • Learn how to implement the Riccati equation in MATLAB for control systems
  • Explore the properties of symmetric matrices in linear algebra
  • Investigate coupled differential equations and their solutions in control theory
USEFUL FOR

Students and professionals in control engineering, applied mathematics, and anyone involved in optimal control system design and analysis.

JavierOlivares
Messages
14
Reaction score
0

Homework Statement



I was wondering if I can get some help on a Linear Regulator Problem for an Optimal Control Problem. Given a state equation and performance measure I am trying to solve using the Riccati equation on MATLAB. This is a sample example I got from a book Optimal Control Donald Kirk. I don't understand how they derived three separate differential equations from the Riccati equation:

The problem goes as followed:

Consider the system https://www.physicsforums.com/attachments/upload_2016-11-26_20-25-28-png.109458/
https://www.physicsforums.com/attachments/upload_2016-11-26_20-26-3-png.109459/

How did they go from a single equation K to three separate equations. I keep looking for resources but many other examples seem to skip this part. Thank for any help or input.

Homework Equations


https://www.physicsforums.com/attachments/upload_2016-11-26_20-26-46-png.109460/

The Attempt at a Solution


[/B]
upload_2016-11-26_20-43-29.png

upload_2016-11-26_20-43-49.png
 
Last edited by a moderator:
Physics news on Phys.org
https://www.physicsforums.com/attachments/upload_2016-11-26_20-39-19-png.109463/

I found a paper on this online that gives somewhat of an example of this problem.
https://www.physicsforums.com/attachments/upload_2016-11-26_20-40-26-png.109465/
https://www.physicsforums.com/attachments/upload_2016-11-26_20-41-4-png.109467/
It seems slightly different though. Sorry for any inconvenience.
 
Last edited by a moderator:
JavierOlivares said:

Homework Statement



I was wondering if I can get some help on a Linear Regulator Problem for an Optimal Control Problem. Given a state equation and performance measure I am trying to solve using the Riccati equation on MATLAB. This is a sample example I got from a book Optimal Control Donald Kirk. I don't understand how they derived three separate differential equations from the Riccati equation:

The problem goes as followed:

Consider the system https://www.physicsforums.com/attachments/upload_2016-11-26_20-25-28-png.109458/
https://www.physicsforums.com/attachments/upload_2016-11-26_20-26-3-png.109459/

How did they go from a single equation K to three separate equations. I keep looking for resources but many other examples seem to skip this part. Thank for any help or input.

Homework Equations


https://www.physicsforums.com/attachments/upload_2016-11-26_20-26-46-png.109460/

The Attempt at a Solution


[/B]
View attachment 109470
View attachment 109471

The differential equations for ##\mathbf{K}## and its transpose ##\mathbf{K}^T## are the same; and (as your attachment in post #2 states), ##\mathbf{K}(t_f) = \mathbf{S}##, where ##\mathbf{S}## is a symmetric matrix. Therefore, the solution ##\mathbf{K}(t)## is a symmetric matrix as well. Now just write the Ricatti equation for the symmetric matrix ##\mathbf{K}## in terms of its components:
$$\mathbf{K} = \pmatrix{K_{11}&K_{12}\\K_{12} & K_{22}} $$
 
Last edited by a moderator:
I'm still a little confused. I understand that the matrix is symmetric. I just don't understand how they have it equal on the LHS a row of three 3x1 differential equations when K seems to be a 2x2. That's where I'm confused. I'm thinking this is some linear algebra property that's going over my head. I don't know if the notes I provided actually answer my question. Thanks for the response

upload_2016-11-27_23-2-37.png
 
JavierOlivares said:
I'm still a little confused. I understand that the matrix is symmetric. I just don't understand how they have it equal on the LHS a row of three 3x1 differential equations when K seems to be a 2x2. That's where I'm confused. I'm thinking this is some linear algebra property that's going over my head. I don't know if the notes I provided actually answer my question. Thanks for the response

View attachment 109555

I have not checked the picture: it is too messy and unstructured. However, both sides of your differential equation are 2x2 matrices, so you get 4 coupled differential equations. Since the matrix is symmetric, only three of the equations are different
 
I think I understand now. I was just confused on multiplying the X Matrix by another 2x2 Matrix. I was thinking the equations would combine X11 + X12 as in the case of a 2x2 and 2x1 but it makes sense now. Thanks.
 

Similar threads

  • · Replies 3 ·
Replies
3
Views
2K
  • · Replies 3 ·
Replies
3
Views
2K
Replies
12
Views
2K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 1 ·
Replies
1
Views
1K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 2 ·
Replies
2
Views
1K
  • · Replies 7 ·
Replies
7
Views
2K
Replies
12
Views
2K
Replies
5
Views
2K